More Coffee, Less Crime The Relationship between ...

More Coffee, Less Crime? The Relationship between Gentrification and Neighborhood Crime Rates in Chicago, 1991 to 2005

Andrew V. Papachristos, Chris M. Smith, Mary L. Scherer, and Melissa A. Fugiero Department of Sociology, University of Massachusetts Amherst

This study examines the relationship between gentrification and neighborhood crime rates by measuring the growth and geographic spread of one of gentrification's most prominent symbols: coffee shops. The annual counts of neighborhood coffee shops provide an on-the-ground measure of a particular form of economic development and changing consumption patterns that tap into central theoretical frames within the gentrification literature. Our analysis augments commonly used Census variables with the annual number of coffee shops in a neighborhood to assess the influence of gentrification on three-year homicide and street robbery counts in Chicago. Longitudinal Poisson regression models with neighborhood fixed effects reveal that gentrification is a racialized process, in which the effect of gentrification on crime is different for White gentrifying neighborhoods than for Black gentrifying neighborhoods. An increasing number of coffee shops in a neighborhood is associated with declining homicide rates for White, Hispanic, and Black neighborhoods; however, an increasing number of coffee shops is associated with increasing street robberies in Black gentrifying neighborhoods.

INTRODUCTION

Gentrification is a process that--for better or worse--changes neighborhoods. New people move into a neighborhood, often displacing existing residents. Older housing stock is renovated, repaired, removed, or reconstructed. And, eventually, commerce also changes as new stores and amenities arise to meet the demands of the shifting population. Whether or not such changes are "good" or "bad" for neighborhood residents is heavily debated by politicians, developers, city planners, and academics alike. On the one hand, research documents the harms of gentrification, which include the displacement of disadvantaged populations and the disruption of social networks (Abu-Lughod, 1999; Rymond-Richmond 2007; Smith 1996). On the other hand, gentrification is also viewed (or utilized) as a way to revitalize a neighborhood's economy, cultural heritage, and social organization (Florida 2002; Freeman 2006).

Correspondence should be addressed to Andrew V. Papachristos, Department of Sociology, University of Massachusetts Amherst, Thompson Hall, 200 Hicks Way, Amherst MA 01003; andrewp@soc.umass.edu.

City & Community 10:3 September 2011 doi: 10.1111/j.1540-6040.2011.01371.x C 2011 American Sociological Association, 1430 K Street NW, Washington, DC 20005

215

CITY & COMMUNITY

One dimension of neighborhood change often associated with gentrification is the reduction of crime. The logic of politicians and city planners is that disadvantaged neighborhoods "upgrade" with the influx of more well-to-do residents and, in so doing, reduce crime. Yet, the empirical relationship between gentrification and crime has produced contradictory findings. In some studies, gentrification is found to increase crime (Covington and Taylor 1989; Lee 2010; Van Wilsem, Wittebrood, and De Graaf 2006); in others, gentrification is linked to decreases in crime (Kreager, Lyons, and Hays 2007). These contradictions are partially explained by wide variation in types of gentrification and criminal activity studied.

The present study examines the relationship between gentrification and neighborhood crime in Chicago from 1991 to 2005. We augment the more commonly used census indicators of gentrification (e.g., median household income) with one of gentrification's most widely recognized symbols: coffee shops. More specifically, we operationalize the spatial distribution of this indicator of gentrification through the annual increase in corporate and noncorporate coffee shops. Measuring the number of coffee shops in a neighborhood has the distinct advantage over the more commonly employed census and survey indicators in that coffee shops provide an on-the-ground and visible manifestation of a particular form of gentrification--the increased presence of an amenity often associated with gentrifiers' lifestyles. Furthermore, this measure also captures the role of corporate and private actors (i.e., coffee shop properties) in the gentrification processes. Using fixed effects longitudinal Poisson regression models, we examine (1) how conventional structural indicators of gentrification and the number of coffee shops relate to changing neighborhood levels of homicide and robbery; and (2) how these effects vary by neighborhood racial composition. We find that gentrification associated with increased amenities such as coffee shops is strongly related to declines in homicide and robbery. This process, however, is racialized, with differing effects for White and Hispanic neighborhoods compared to Black neighborhoods.

The paper proceeds as follows. We begin by defining gentrification and reviewing the empirical research on the relationship between gentrification and crime. We situate this project in the larger debates over whether gentrification is a "misunderstood savior" or a "vengeful wrecker" (Atkinson 2003b); in other words, whether gentrification affects the residents of a poor neighborhood for the better or the worse (see Brown-Saracino, 2010).1 Our analysis is presented in three stages: (1) descriptive analysis of the distribution of crime and coffee shops in Chicago; (2) longitudinal analysis of neighborhood levels of gentrification predicting crime; and (3) an analysis of these findings by neighborhood racial composition. The paper concludes with a review of the findings and implications for future research.

GENTRIFICATION AND CRIME

Broadly defined, gentrification is a process that changes the character and composition of a neighborhood, resulting in the direct and indirect displacement of lower income households with higher income households (Clay 1979; Glass 1964; Kennedy and Leonard 2001; Wyly and Hammel 2005; Zukin 1987). For our purposes, we conceptualize gentrification as a churning process that involves the in-migration of wealth and the outmigration of poverty, most often resulting in over time increases in median household

216

MORE COFFEE, LESS CRIME?

incomes, property values, and presence of lifestyle amenities that appeal to the tastes-- and meet the demands of--the wealthier residents (Lees, Slater, and Wyly 2008). The exact definition of "gentrification"--not to mention its political utility--stirs much academic debate (see Brown-Saracino 2010; Freeman 2005). To be clear, our study does not seek to redefine gentrification. Instead, we contribute an additional indicator of gentrification that follows from qualitative research and that may have important implications for neighborhood-level quantitative research. However, this study does address an important debate within the gentrification literature--namely, whether gentrification is "good" or "bad" for neighborhoods, residents, and the larger urban environment.

While academics have mainly debated the implications of gentrification for theories of social change, urban demography, and inequality, activists, politicians, and the general public are divided over whether gentrification is beneficial or detrimental for neighborhood residents (Atkinson 2003b, 2004; Brown-Saracino 2010). On the one hand, increased tax revenue, improved schools, and neighborhood investments associated with gentrification are potentially beneficial outcomes for poorer residents (Florida 2002; Freeman 2006). A study by Freeman (2006), for example, suggests that longtime residents can benefit from new social connections to in-movers' social, cultural, and economic capital, which enables them to demand better police patrol after dark, send their children to improving schools, and receive other public services. Likewise, Pattillo's (2007) work on Black gentrifiers found that some longtime residents believed that gentrification (at least in its early stages) improved their quality of life. On the other hand, displacement, increased rents, and new forms of surveillance are potentially detrimental--if not outright harmful--outcomes of gentrification for longtime residents (Chernoff 1980; Levy and Cybriwsky 1980; Pattillo 2007; Perez 2004). Indeed, the disruption of social support networks and local neighborhood social organization in disadvantaged communities is one of the greatest casualties of gentrification (Rymond-Richmond 2007). Although what constitutes the benefits and detriments of gentrification varies across social groups and over time (Kerstein 1990; Lloyd 2005), in most cases, the gentrifiers benefit while the gentrified suffer.

This debate over the consequences of gentrification frequently invokes crime rates. The implicit connection between gentrification and crime relates to the ways that gentrification processes might alter the neighborhood conditions associated with crime. Proponents of gentrification widely claim (though rarely empirically assess) that reductions in neighborhood crime rates are a benefit of gentrification at the neighborhood and even city level. This logic assumes that the influx of more affluent residents and improvements in resources, institutions, and amenities lower crime rates and, thereby, improve overall safety in gentrifying neighborhoods (McDonald 1986). Another argument posits that gentrification ameliorates crime and delinquency through increased law enforcement efforts, new economic and social opportunities, or the displacement of prior criminal residents (see Kirk and Laub 2010).2

Despite the frequency with which the literature mentions the relationship between gentrification and crime, there are surprisingly few empirical tests of this relationship. Likewise, criminologists studying neighborhood change frequently raise the issue of gentrification as an important social process but rarely measure it. The research that does test this relationship has produced evidence of both a negative and a positive association between gentrification and crime.

217

CITY & COMMUNITY

Several quantitative studies have found a positive relationship between gentrification and crime--crime actually increases as neighborhoods gentrify (Covington and Taylor 1989; Lee 2010; Taylor and Covington 1988; Van Wilsem et al. 2006). As such, gentrification might be considered "bad" for the overall safety of neighborhood residents. Routine activities and rational choice theories suggest crime arrives as the result of the convergence in time and space of motivated offenders, lack of social control, and suitable targets (Felson 1994); as such, the in-migration of wealth into poorer areas may create new and more lucrative opportunities for particular types of crimes, especially property and economic crimes. Social disorganization theory suggests that gentrifying neighborhoods might experience an increase in crime rates as the neighborhood social structures undergo a period of flux and socioeconomic heterogeneity, which lessen a community's ability to control crime internally. Yet, as the community continues to gentrify and population changes stabilize, disorder should decline and social organization should increase (Kirk and Laub 2010). For instance, as neighborhoods stabilize, coffee shops (like other social institutions) might provide additional "eyes on the streets" to monitor public behavior. Crime, therefore, would eventually decrease as neighborhood structural conditions stabilize, thereby improving the overall safety of neighborhood residents. Along these lines, Kreager et al. (2007) found a curvilinear relationship between gentrification and property crime in Seattle, with an increasing and positive relationship when the gentrification process first begins, followed by a decline in property crimes as gentrification progresses; this study also finds an overall negative effect of gentrification on violent crime (Kreager et al. 2007). Lee (2010) also finds both positive and negative effects in his study of gentrification in Los Angeles that varied by the type of crime and neighborhood examined; however, the majority of his crime outcomes (seven out of 11) did not yield significant results.

Recent developments in the study of neighborhoods and crime might help to unpack such divergent findings, especially as they pertain to racial disparities in rates of crime and violence. This line of inquiry focuses on how neighborhood-level social processes influence outcomes such as health, political participation, and levels of crime and violence (Kirk and Laub 2010; Sampson, Morenoff, and Gannon-Rowley 2002). Most relevant for the present study, this body of work consistently demonstrates that neighborhood structural characteristics, such as concentrated disadvantage, tend to be the largest and most resilient predictors of neighborhood crime and violence regardless of the racial composition of a neighborhood (Peterson and Krivo 1999; Pratt and Cullen 2005). Furthermore, and again regardless of the racial composition of a neighborhood, higher levels of trust, social cohesion, and informal social organization among neighborhood residents are associated with lower levels of crime and delinquency (Kirk and Papachristos 2011; Sampson, Raudenbush, and Earls 1997).

Mounting evidence further supports the fact that racial disparities in crime tend to be confounded with significant differences in community contexts (Sampson and Bean 2006; Sampson and Wilson 1995). Often referred to as "ecological dissimilarity," this research posits that it is impossible for statistical models to reproduce the same neighborhood conditions for Whites and Blacks, since they occupy markedly different positions in the social and spatial order of the city (Sampson and Bean 2006). For example, a recent study by Kirk and Papachristos (2011) finds that Bronzeville, a Black gentrifying neighborhood in Chicago with considerable levels of social control, continues to have high crime rates, in part because it is geographically surrounded by other

218

MORE COFFEE, LESS CRIME?

high-crime/low-social control disadvantaged neighborhoods. In contrast, White neighborhoods with comparable levels of social control experience greater crime declines because they are spatially adjacent to neighborhoods with higher levels of social control (see also Morenoff, Sampson, and Raudenbush 2001). In short, even though informal social control is associated with lower levels of crime, regardless of race, such processes unfold differently given the dissimilarity in the ecological conditions between White and Black neighborhoods.

We maintain that gentrification may similarly vary by the racial composition of a neighborhood, the racial composition of the gentrifiers, and the ecological position of a neighborhood in a city. The ecological dissimilarity hypothesis would suggest that Whites, Blacks, and Hispanics gentrify and experience gentrification in different ways, not just because of individual characteristics and preferences, but also because of the community contexts in which they reside or the neighborhoods they gentrify. In fact, prior research already suggests that gentrification is a highly racialized process. In the 1960s and 1970s, most gentrifiers tended to be college-educated, younger Whites who moved into workingclass White and non-White Hispanic communities (Clay 1979). This continues today, as in the case of Brooklyn's Williamsburg, a formerly Jewish and Hispanic working-class neighborhood that is now associated with a White, young, single, educated (but nonwealthy), artsy crowd called "hipsters"--the modern equivalent of "bohemians" (Zukin 2010). Historically, Black Brooklyn neighborhoods, such as Bedford-Stuyvesant, are associated with far less displacement of long-time residents (Freeman 2006) and, thus, are more likely to remain Black even though the economic and cultural organization of the neighborhood may change. Gentrifiers in these neighborhoods tend to be members of the Black middle class or "buppies"--Black Urban Professionals (Hyra 2008; Pattillo 2007).

However, whether or not the effect of gentrification on other neighborhood outcomes--such as crime--varies by the racial composition of a neighborhood is not well understood. The research on gentrification and crime has not fully considered how gentrification might vary by the racial composition of a neighborhood, nor has the criminological research on the subject fully measured gentrification processes. Yet, given the observed disparities of crime rates along racial and ethnic dimensions, the effects of gentrification may also vary by the racial composition and ecological position of a neighborhood. Gentrification may unfold differently in Black, Hispanic, and White neighborhoods based on individual preferences, the practices of corporations and banks (Wacquant 1989; Yinger 1995), and the distribution of amenities and retail establishments across socioeconomic neighborhood conditions (Small and McDermott 2006).

The present study contributes to the gentrification and crime debate by improving how neighborhood scholars measure gentrification. Specifically, we devise a noncensus indicator identified in qualitative research on gentrification: the number of coffee shops in a neighborhood. Whereas census indicators provide measurements of demographic and economic conditions, our measure of coffee shops encompasses additional aspects of gentrification such as lifestyle, consumption patterns, and the capacity of corporate and private actors. Consistent with prior research on neighborhoods and crime, we hypothesize that crime rates will decline at a greater rate in gentrifying neighborhoods as population shifts stabilize. However, consistent with the ecological dissimilarity framework, we suspect that the effect of gentrification on crime will vary by neighborhood racial composition. More specifically, any crime-reducing effect associated with gentrification will most likely be lower in Black neighborhoods as compared to non-Black neighborhoods.

219

CITY & COMMUNITY

INDICATORS OF GENTRIFICATION AND CRIME

Gentrification is a process and, therefore, necessitates a measure that increases or decreases quantitatively to enable meaningful comparisons over time and across neighborhoods. Yet, gentrification is a multifaceted, complex process involving political, corporate, and independent actors. Furthermore, gentrification often unfolds at an uneven temporal pace, sometimes occurring quickly (e.g., the demolition of public housing) and other times occurring more slowly (e.g., turnover of neighborhood population). Hence, gentrification is not necessarily linear or cumulative. Given this complexity, qualitative methods are perhaps best suited to study gentrification as a contextual experience and a powerful generator of urban inequality. Nevertheless, a quantitative indicator of gentrification is necessary if we wish to test the popularly and politically employed theory that gentrification leads to improved neighborhoods regardless of racial composition and time period.

To date, quantitative gentrification measures have relied almost exclusively on census data. The resulting research is often limited by inadequate measures of displacement (Hammel and Wyly 1996; Kreager et al. 2007; Taylor and Covington 1988), cross-sectional measures of gentrification (Kreager et al. 2007; Wyly and Hammel 19993), or measures that employ limited definitions of gentrification such as only changes in residential income (Lee 2010). Given this lack of attention to adequate quantitative measures of gentrification, it is little surprise that the research on gentrification and crime has produced divergent results.

While census-based indicators of gentrification are able to capture some important dimensions of gentrification, such indicators face at least two significant limitations. First, the decennial availability of the census means that measures of neighborhood change must be interpolated using 10-year time intervals, essentially estimating any change associated with gentrification as a linear trend. Second, relying on aggregate neighborhood level, census indicators assume that only individual residents drive gentrification processes, an assumption that fails to consider of the role of corporate and political actors in the gentrification process.

The goal of our study is to augment commonly used census indicators of gentrification with a noncensus-based measure that maps onto the nonlinear tendencies of gentrification and that can be measured precisely over time and space--the number of coffee shops in a neighborhood. We have chosen coffee shops as an indicator of gentrification because of their prominent place in both the urban imaginary (Atkinson 2003a; Lloyd 2005; Zukin 2008) and in policy prescriptions from consultants (see Florida 2002).

Prior qualitative gentrification scholarship refers to the appearance of coffee shops as a meaningful representation of neighborhood change. For example, urban sociologist Richard Lloyd traces the life and death of Urbus Orbis, a legendary punk-rock coffee shop in the gentrified Chicago neighborhood of Wicker Park. Lloyd (2005) suggests that Urbus Orbis' magnetism in the 1990s, its reputation as a gathering spot for newcomers, its eventual status as an iconic outpost of gentrification in the once poor Hispanic neighborhood, and its closing shortly before the neighborhood's feature in the television series The Real World are all emblematic of a coffee shop's role in neighborhood change. Although rising commercial rents and financial viability ushered in the death of Urbus Orbis, a Starbucks currently stands less than 400 feet from the old Urbus, and four additional Starbucks and eight other independent coffee shops can be found within a

220

MORE COFFEE, LESS CRIME?

one-mile radius. In much the same way, the presence of coffee shops, specifically Starbucks, is referenced in other qualitative and descriptive accounts of gentrified neighborhoods (e.g. Boyd 2008; Hyra 2008; Kennedy and Leonard 2001; Wyly and Hammel 2005), but to the best of our knowledge, not a single quantitative study has actually measured the temporal and geographic distribution of coffee shops and analyzed their influence on other neighborhood processes.

There are also theoretical reasons to use coffee shops as indicators of the neighborhood shifts associated with gentrification.4 When scholars debate whether the engine of gentrification is economic or cultural, they are asking whether something like a proliferation of consumption amenities in a previously disinvested area is a response to supply or demand (Brown-Saracino 2010). Coffee shops, however, are clearly both. While they do require some start-up capital and a market, cafe?s and coffee shops have for centuries been used as "third places," serving the critical function in the community as locations where people may socialize and retreat from home and work (Oldenburg 1999). Though this function is less applicable to the contemporary United States (many coffee shops today are primarily a modern convenience for the on-the-go caffeine consumer), the commodification of third place nostalgia has proven a successful marketing tool for corporate coffee shops such as Starbucks (Simon 2009).5 Coffee sellers use specific marketing language to recreate high-culture ideas tied to art and philosophy for its customers, targeting an ideal bourgeois patron (Roseberry 1996; Simon 2009; Thompson and Arsel 2004). Not dealing in a necessary comestible product, such as milk or bread, but rather a status product, coffee shops are integral to the leisure and lifestyle amenities package so attractive to urban gentrifiers. In a postneed economy, coffee shops meet the urban consumer's demands for a space to meet friends or use the Internet, demands which were mostly absent from the neighborhood's prior population.

In short, we argue that measuring the number of coffee shops located in a neighborhood each year provides an almost real-time measure of one type of development commonly associated with gentrification. To this end, we counted the total number of coffee shops in a neighborhood each year from 1991 to 2004 from all listings under "Coffee & Tea" and "Coffee Shops" in annual Chicago Business Directories. DirectoriesUSA compiles these data from public records including phone books, annual reports, courthouse filings, etc.6 Recent research suggests that business directories and geocoding business addresses are a resilient and robust way to measure urban change (Bader et al. 2010; Carroll and Torfason 2011; Kubrin et al. 2011; Schlichtman and Patch 2008; Small and McDermott 2006). Our final coffee shop variable is constructed as a three-year average count of coffee shops for the 15-year period between 1991 and 2005, yielding a total of five time periods.

DATA

The sources of the other data utilized in this study are the U.S. Census and the Chicago Police Department (CPD). The units of analysis used here are 341 neighborhood clusters over time identified in prior research by Sampson et al. (1997).7 The Project on Human Development in Chicago Neighborhoods created these neighborhood clusters by combining the 847 census tracts of Chicago into geographically contiguous areas that are internally homogenous on the key census indictors of race/ethnicity, housing

221

CITY & COMMUNITY

density, and family structure (Sampson et al. 1997: 919). Though not without drawbacks, the use of neighborhood clusters as the unit of analysis has two major benefits over census boundaries. First, previous research demonstrates that these neighborhood clusters are ecologically meaningful (Sampson et al. 1997). Second, given the rarity of crime as an event and the even smaller number of coffee shops in Chicago, aggregating to the neighborhood cluster reduces the number of zeros used in the analysis and, therefore, minimizes some of the overdispersion in the distribution of key variables.

DEPENDENT VARIABLES

Our empirical analysis examines the temporal diffusion of gentrification across neighborhood clusters' levels of crime. The main dependent variables are the annual counts of homicides and street robberies.8 Homicide and robbery tend to function as strong indicators of the overall level of neighborhood violent crime. Homicide data, in particular, have several advantages over the use of other crime measures because (1) there is a close match between known homicides and the true number of homicides--that is, homicides are highly likely to be reported or discovered by police, and (2) homicide is less susceptible to definitional variation by the police. The threat of robbery victimization has far-reaching effects on urban life through its influence on choices for residents and visitors about where to live, work, shop, and dine; as such, robbery operates as a measure of how "safe" a neighborhood feels to residents. Because crime is a rare event, it is standard practice to construct counts based on three-year periods (e.g., Morenoff et al. 2001). These three-year periods also reduce the number of zeros in the dependent variables. For the present analyses, we created a measure of three-year counts of homicide and street robberies for each of the 341 neighborhoods in Chicago for the 15-year period between 1991 and 2005, yielding a total of five time periods: 1991?1993, 1994?1996, 1997?1999, 2000?2002, and 2003?2005.

NEIGHBORHOOD STRUCTURAL CHARACTERISTICS: A FACTOR ANALYTIC APPROACH

Quantitative research on gentrification tends to include an extensive list of census-based measures of levels of education, income, residence costs, kinds of occupations, and other population characteristics (Hammel and Wyly 1996). Keeping with this tradition, our analysis includes seven census indicators: percent of population with a bachelor's degree, percent of population that moved into the neighborhood in the last five years, percent of new housing built in last five years, the log of mean family income, the percent of population that is Black, the percent of population that is Hispanic, and the percent of population that is foreign born. These individual census indicators are highly correlated with themselves across time, suggesting that very few neighborhoods experience any detectable change along these dimensions (see also Sampson 2009).9 Thus, while gentrification can have dramatic consequences, it is quite a rare phenomenon when measured using only census indicators. Our measure of coffee shops is specifically designed to try to capture a more subtle cultural process of neighborhood change that might not be captured by such census indicators.

222

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download